Re: [PATCH] mm: mmap_lock: optimize mmap_lock tracepoints

[Shakeel Butt <shakeel.butt@linux.dev>]

On Fri, Nov 22, 2024 at 10:46:53PM -0800, Yosry Ahmed wrote:
> On Fri, Nov 22, 2024 at 10:10 PM Shakeel Butt <shakeel.butt@linux.dev> wrote:
> >
> > We are starting to deploy mmap_lock tracepoint monitoring across our
> > fleet and the early results showed that these tracepoints are consuming
> > significant amount of CPUs in kernfs_path_from_node when enabled.
> >
> > It seems like the kernel is trying to resolved the cgroup path in the
> 
> s/resolved/resolve
> 
> > fast path of the locking code path when the tracepoints are enabled. In
> > addition for some application their metrics are regressing when
> > monitoring is enabled.
> >
> > The cgroup path resolution can be slow and should not be done in the
> > fast path. Most userspace tools, like bpftrace, provides functionality
> > to get the cgroup path from cgroup id, so let's just trace the cgroup
> > id and the users can use better tools to get the path in the slow path.
> >
> > Signed-off-by: Shakeel Butt <shakeel.butt@linux.dev>
> > ---
> >  include/linux/memcontrol.h       | 18 ++++++++++++
> >  include/trace/events/mmap_lock.h | 32 ++++++++++----------
> >  mm/mmap_lock.c                   | 50 ++------------------------------
> >  3 files changed, 36 insertions(+), 64 deletions(-)
> >
> > diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
> > index 5502aa8e138e..d82f08cd70cd 100644
> > --- a/include/linux/memcontrol.h
> > +++ b/include/linux/memcontrol.h
> > @@ -1046,6 +1046,19 @@ static inline void memcg_memory_event_mm(struct mm_struct *mm,
> >
> >  void split_page_memcg(struct page *head, int old_order, int new_order);
> >
> > +static inline u64 memcg_id_from_mm(struct mm_struct *mm)
> 
> The usage of memcg_id here and throughout the patch is a bit confusing
> because we have a member called 'id' in struct mem_cgroup, but this
> isn't it. This is the cgroup_id of the memcg. I admit it's hard to
> distinguish them during naming, but when I first saw the function I
> thought it was returning memcg->id.
> 
> Maybe just cgroup_id_from_mm()? In cgroup v2, the cgroup id is the
> same regardless of the controller anyway, in cgroup v1, it's kinda
> natural that we return the cgroup id of the memcg.
> 
> I don't feel strongly, but I prefer that we use clearer naming, and
> either way a comment may help clarify things.
> 

Ack, I will change to cgroup_id_from_mm() but I will keep memcg_id in
the tracepoints.

> > +{
> > +       struct mem_cgroup *memcg;
> > +       u64 id = 0;
> > +
> > +       rcu_read_lock();
> > +       memcg = mem_cgroup_from_task(rcu_dereference(mm->owner));
> > +       if (likely(memcg))
> > +               id = cgroup_id(memcg->css.cgroup);
> 
> We return 0 if the memcg is NULL here, shouldn't we return the cgroup
> id of the root memcg instead? This is more consistent with
> get_mem_cgroup_from_mm(), and makes sure we always return the id of a
> valid cgroup.

Good point and I need to add a mem_cgroup_disabled() check as well. Will
do in v2.